|
1
|
|
|
#!/usr/bin/env python |
|
2
|
|
|
# -*- coding: utf-8 -*- |
|
3
|
|
|
""" |
|
4
|
|
|
A set of dataclasses concerning roles of persons and their particulars. |
|
5
|
|
|
""" |
|
6
|
|
|
import os |
|
7
|
|
|
import sys |
|
8
|
|
|
from dataclasses import dataclass, field |
|
9
|
|
|
from typing import List, Optional |
|
10
|
|
|
|
|
11
|
|
|
PACKAGE_PARENT = ".." |
|
12
|
|
|
SCRIPT_DIR = os.path.dirname( |
|
13
|
|
|
os.path.realpath(os.path.join(os.getcwd(), os.path.expanduser(__file__))) |
|
14
|
|
|
) # isort:skip # noqa # pylint: disable=wrong-import-position |
|
15
|
|
|
sys.path.append( |
|
16
|
|
|
os.path.normpath(os.path.join(SCRIPT_DIR, PACKAGE_PARENT)) |
|
17
|
|
|
) # isort: skip # noqa # pylint: disable=wrong-import-position |
|
18
|
|
|
|
|
19
|
|
|
from person import Person # type: ignore # noqa |
|
20
|
|
|
from src.resources import helpers # type: ignore # noqa |
|
21
|
|
|
from src.resources.constants import GERMAN_PARTIES # type: ignore # noqa |
|
22
|
|
|
from src.resources.constants import PEER_PREPOSITIONS # type: ignore # noqa |
|
23
|
|
|
from src.resources.constants import PEERTITLES # type: ignore # noqa |
|
24
|
|
|
from src.resources.helpers import ( # type: ignore # noqa |
|
25
|
|
|
AttrDisplay, |
|
26
|
|
|
NotInRange, |
|
27
|
|
|
Party, |
|
28
|
|
|
TooManyFirstNames, |
|
29
|
|
|
) |
|
30
|
|
|
|
|
31
|
|
|
|
|
32
|
|
|
@dataclass |
|
33
|
|
|
class _Politician_default: |
|
34
|
|
|
electoral_ward: str = field(default="ew") |
|
35
|
|
|
ward_no: Optional[int] = field(default=None) |
|
36
|
|
|
voter_count: Optional[int] = field(default=None) |
|
37
|
|
|
minister: Optional[str] = field(default=None) |
|
38
|
|
|
offices: List[str] = field(default_factory=lambda: []) |
|
39
|
|
|
parties: List[str] = field(default_factory=lambda: []) |
|
40
|
|
|
|
|
41
|
|
|
def renamed_wards(self): |
|
42
|
|
|
wards = { |
|
43
|
|
|
"Kreis Aachen I": "Aachen III", |
|
44
|
|
|
"Hochsauerlandkreis II – Soest III": "Hochsauerlandkreis II", |
|
45
|
|
|
"Kreis Aachen II": "Aachen IV" |
|
46
|
|
|
if self.last_name in ["Wirtz", "Weidenhaupt"] |
|
47
|
|
|
else "Kreis Aachen I", |
|
48
|
|
|
} |
|
49
|
|
|
if self.electoral_ward in wards.keys(): |
|
50
|
|
|
self.electoral_ward = wards[self.electoral_ward] |
|
51
|
|
|
|
|
52
|
|
|
def scrape_wiki_for_ward(self) -> None: |
|
53
|
|
|
import requests |
|
54
|
|
|
from bs4 import BeautifulSoup # type: ignore |
|
55
|
|
|
|
|
56
|
|
|
URL_base = "https://de.wikipedia.org/wiki/Landtagswahlkreis_{}" |
|
57
|
|
|
URL = URL_base.format(self.electoral_ward) |
|
58
|
|
|
req = requests.get(URL) |
|
59
|
|
|
bsObj = BeautifulSoup(req.text, "lxml") |
|
60
|
|
|
table = bsObj.find(class_="infobox float-right toptextcells") |
|
61
|
|
|
self.scrape_wiki_table_for_ward(table) |
|
62
|
|
|
|
|
63
|
|
|
def scrape_wiki_table_for_ward(self, table) -> None: |
|
64
|
|
|
for td in table.find_all("td"): |
|
65
|
|
|
if "Wahlkreisnummer" in td.text: |
|
66
|
|
|
ward_no = td.find_next().text.strip() |
|
67
|
|
|
ward_no = ward_no.split(" ")[0] |
|
68
|
|
|
self.ward_no = int(ward_no) |
|
69
|
|
|
elif "Wahlberechtigte" in td.text: |
|
70
|
|
|
voter_count = td.find_next().text.strip() |
|
71
|
|
|
voter_count = self.fix_voter_count(voter_count) |
|
72
|
|
|
self.voter_count = int(voter_count) |
|
73
|
|
|
|
|
74
|
|
|
def fix_voter_count(self, voter_count): |
|
75
|
|
|
if voter_count[-1] == "]": |
|
76
|
|
|
voter_count = voter_count[:-3] |
|
77
|
|
|
if " " in voter_count: |
|
78
|
|
|
voter_count = "".join(voter_count.split(" ")) |
|
79
|
|
|
else: |
|
80
|
|
|
voter_count = "".join(voter_count.split(".")) |
|
81
|
|
|
return voter_count |
|
82
|
|
|
|
|
83
|
|
|
|
|
84
|
|
|
@dataclass |
|
85
|
|
|
class Politician( |
|
86
|
|
|
_Politician_default, |
|
87
|
|
|
helpers._Party_default, |
|
88
|
|
|
Person, |
|
89
|
|
|
helpers._Party_base, |
|
90
|
|
|
AttrDisplay, |
|
91
|
|
|
): |
|
92
|
|
|
def __post_init__(self): |
|
93
|
|
|
Party.__post_init__(self) |
|
94
|
|
|
Person.__post_init__(self) |
|
95
|
|
|
Person.get_sex(self) |
|
96
|
|
|
Person.get_age(self) |
|
97
|
|
|
self.change_ward() |
|
98
|
|
|
if self.party_name in GERMAN_PARTIES: |
|
99
|
|
|
self.parties.append( |
|
100
|
|
|
Party(self.party_name, self.party_entry, self.party_exit) |
|
101
|
|
|
) |
|
102
|
|
|
if self.minister and self.minister not in self.offices: |
|
103
|
|
|
self.offices.append(self.minister) |
|
104
|
|
|
|
|
105
|
|
|
def add_Party( |
|
106
|
|
|
self, party_name, party_entry="unknown", party_exit="unknown" |
|
107
|
|
|
): # noqa |
|
108
|
|
|
if party_name in GERMAN_PARTIES: |
|
109
|
|
|
if self.party_is_in_parties(party_name, party_entry, party_exit): |
|
110
|
|
|
pass |
|
111
|
|
|
else: |
|
112
|
|
|
self.parties.append(Party(party_name, party_entry, party_exit)) |
|
113
|
|
|
self.party_name = party_name |
|
114
|
|
|
self.party_entry = party_entry |
|
115
|
|
|
self.party_exit = party_exit |
|
116
|
|
|
|
|
117
|
|
|
def align_party_entries( |
|
118
|
|
|
self, party, party_name, party_entry, party_exit |
|
119
|
|
|
) -> Party: # noqa |
|
120
|
|
|
if party_entry != "unknown" and party.party_entry == "unknown": |
|
121
|
|
|
party.party_entry = party_entry |
|
122
|
|
|
if party_exit != "unknown" and party.party_exit == "unknown": |
|
123
|
|
|
party.party_exit = party_exit |
|
124
|
|
|
return party |
|
125
|
|
|
|
|
126
|
|
|
def party_is_in_parties(self, party_name, party_entry, party_exit): |
|
127
|
|
|
parties_tmp = self.parties[:] |
|
128
|
|
|
for party in parties_tmp: |
|
129
|
|
|
if party_name == party.party_name: |
|
130
|
|
|
party_updated = self.align_party_entries( |
|
131
|
|
|
party, party_name, party_entry, party_exit |
|
132
|
|
|
) |
|
133
|
|
|
self.parties.remove(party) |
|
134
|
|
|
self.parties.append(party_updated) |
|
135
|
|
|
self.party_entry = party_updated.party_entry |
|
136
|
|
|
self.party_exit = party_updated.party_exit |
|
137
|
|
|
return True |
|
138
|
|
|
return False |
|
139
|
|
|
|
|
140
|
|
|
def change_ward(self, ward=None): |
|
141
|
|
|
if ward: |
|
142
|
|
|
self.electoral_ward = ward |
|
143
|
|
|
if self.electoral_ward not in ["ew", "Landesliste"]: |
|
144
|
|
|
self.renamed_wards() |
|
145
|
|
|
self.scrape_wiki_for_ward() |
|
146
|
|
|
else: |
|
147
|
|
|
self.electoral_ward = "ew" |
|
148
|
|
|
|
|
149
|
|
|
|
|
150
|
|
|
if __name__ == "__main__": |
|
151
|
|
|
|
|
152
|
|
|
politician = Politician( |
|
153
|
|
|
"SPD", |
|
154
|
|
|
"Bärbel", |
|
155
|
|
|
"Gutherz", |
|
156
|
|
|
academic_title="Dr.", |
|
157
|
|
|
date_of_birth="1980", |
|
158
|
|
|
electoral_ward="Köln I", |
|
159
|
|
|
) |
|
160
|
|
|
print(politician) |
|
161
|
|
|
|
0LL13 /
person
